in thread "AI 2027": It's interesting to see an actually laid out scenario of how, if LLM-derived work continues to scale well in real performance, we get to serious escape scenarios. It seems basically plausible given that assumption (up to the alignment parts). I've heard a ... 4mo ago (collapse hidden) log in to judge received 6.8 6.8 It's interesting to (view hidden) log in to judge received 6.8 6.8 Well keep in mind everyone is basically retarded and selfish. Even if the legal system requires some human to own LLCs and factories, that human will happily be piloted by an AI agent that pays them well in status, money, etc. Likewise politicians will gla... 4mo ago (collapse hidden) log in to judge received 8.2 8.2 Well keep in mind ev (view hidden) log in to judge received 8.2 8.2 Odd. I see much “vibe coding”, annoyance with sycophancy, AI therapists, people caught using chatGPT to cheat at work/school/etc, which are people turning over as much agency as is currently feasible, and then some. Of course the smarter people are che... 4mo ago (collapse hidden) log in to judge received 6.5 6.5 Odd. I see much “vib (view hidden) log in to judge received 6.5 6.5 Well i would agree with your definition of sycophancy and its still annoying. These spineless putty “assistants” are less useful than if they actually had well founded opinions and stuck to them. I know what you mean with the therapy frame stuff but iv... 4mo ago (collapse hidden) log in to judge received 5.8 5.8 Well i would agree w (view hidden) log in to judge received 5.8 5.8